A High-Risk Subgroup Multiple Myeloma Classification Based on the Detection of PR Minor Subclones

In the last two decades several classifications of Multiple Myeloma (MM) have been proposed based on gene expression signatures. These include the supervised translocation and cyclin D (TC) molecular subgroup approach or unsupervised methods to elucidate the molecular heterogeneity of MM and identify high risk subgroups. In particular, two RNA based gene expression studies have identified 8 to 10 unique subtypes, including the MS, MAF, CD1, CD2, and PR subtypes. The PR subgroup represents a highly proliferative gene signature that is associated with very poor survival outcomes. However, assigning this PR subgroup based on bulk transcriptome signatures can be challenging giving the mixture of genetic backgrounds that can be observed within this subtype. Furthermore, the presence of a minor PR subclone can often be missed with bulk genomic studies leading to the misclassification of a patient as standard risk. We here postulated that minor PR subclones expand under selective therapeutic pressures to become the predominant clones at subsequent relapses and the presence of PR subclones can impact any of the molecular subgroups pertaining poor survival outcomes.

Bulk RNA or single cell RNA (scRNAseq) analysis were performed in bone marrow sorted primary MM cells obtained from 718 patients from the CoMMpass trial and 32 relapse/refractory MM patients, respectively. For scRNAseq, serial samples with paired baseline and relapse samples were available in 18 cases. Bone marrow plasma cells samples were separated from bone marrow aspirates using positive sorting selection of CD138⁺ cells. For bulk RNA-seq we used the mRNA HyperPrep kit (Kapa Biosystems) with RNA from 50,000 cells. For scRNA studies unbiased mRNA profiling was conducted using the GemCode system (10x Genomics). Cell Ranger and Seurat were used for sample de-multiplexing, barcode processing, single-cell 3' gene and data analysis. Sequencing was performed on Illumina. By performing GSEA at the single cell level using the Zhan gene data set we have classified patients in the appropriate TC groups. FISH data was used to insure accurate calls. For the survival analysis OS was analyzed by the Kaplan-Meier method.

In the CoMMpass dataset, with bulk RNA sequencing, the PR subtype contained 7.1% of patients with an admixture of classic genetic subtypes and very poor clinical outcome, with a median OS of 21.3 months. High proliferation index scores were concentrated in this subtype. There was also a strong tendency for patients to transition to the PR subtype at progression, with 25.5% of patients in a non-PR subtype at diagnosis transitioning to PR. Regardless of original subtype, patients that transitioned to the PR subtype rapidly succumbed to their disease with a median OS after the detected progression of 88 days. Using subtype prediction weight we could identify non-PR patients with PR subtype weights above 5% and these patients have a significantly shorter OS of 50 months compared to patients with <5% PR.

At the single cell level, investigating the relapsed patients cohort, we have analyzed 92,139 cells and identified 59 clusters. A PR signature (defined as present in ≥15% of cells) was identified in 38% of cases despite baseline favorable prognostic group by FISH cytogenetics (7% HP, 28% t(11;14), and 57% normal FISH). A good concordance with FISH data was observed. In serially relapsed paired samples, all the PR patients maintain their signature overtime and 44% of patients transitioned to PR at subsequent relapse (Fig1A). Survival analyses were performed to evaluate how the presence of minor PR clones at the single cell level relate to clinical outcomes. Notably, the overall survival of patients classified as PR was dramatically reduced with a median OS of 6 months compared to 34 months for non-PR patients (Fig1B).

In conclusion, we here show that minor PR clones defined by transcriptional signature through bulk or scRNA may be detectable in patients with variable MM molecular subgroups, including those classified as standard risk by the currently adopted classifications. We also show that the presence of these PR minor clones pertains adverse outcomes with very poor survivals. At relapse, most acquired PR cases originated from standard-risk presentations consistent with a Darwinian clonal evolution. Future classifications should account for the presence of these PR subclones for a better disease prognostication and developments of novel therapeutics.